Electric Automation Forum
Forum » General Discussion » Phase shifted full bridge converter dead time
Topics: Phase shifted full bridge converter dead time on General Discussion
#1
Start by
venkata subash
11-21-2013 02:09 AM

Phase shifted full bridge converter dead time

what happens if the dead time is very much higher than ZVS transition time in phase shifted full bridge converter IGBT as switch
11-21-2013 04:10 AM
Top #2
Ray Ridley
11-21-2013 04:10 AM
It would seem to me that eventually the converter comes out of the diode conduction phase and if you are unlucky with the exact sequence of events, you will force-commutate the weak diode and get into trouble.

But not always. That is the big and often elusive failure in these converters.
11-21-2013 06:44 AM
Top #3
Joseph Youssef
11-21-2013 06:44 AM
the Vds will oscilates back high, then it is not ZVS any more, more losses and radiated emission.
Also, at higher frequency, a portion of your duty cycle will be lost and lead to a limited duty cycle., which can force you to change the turn ration on the transformer.
11-21-2013 09:41 AM
Top #4
Darrell Hambley P.E.
11-21-2013 09:41 AM
Venkata, If the dead time is long enough so that the current through the reverse diodes stops, a resonant action due to the series inductance and your capacitance at the junction will ring to some other value of voltage which is not the "zero volts" which you want for the ZVS. When the turn on of the IGBT finally occurs, a 1/2*C*V^2 loss will happen.
11-21-2013 11:55 AM
Top #5
Mau Pham
11-21-2013 11:55 AM
Darrell:

So if the dead time is slightly too long, you'll dissipate an energy of 1/2*C*V^2 every time you turn on a power transistor.

But if the Drain Source voltage of the transistor is less than a couple of volts (compared to the supply voltage of hundreds of volts), that energy loss is tiny with respect to the power delivered to the load. So you'll lose efficiency by less than a couple percentages.

On the other hand, if the dead time is slightly too short, you'll draw sharp spikes of current through the input supply, causing supply noise as well as energy loss. Possibly EMI and RFI. Also a possibility of hard damage.

Since it might be difficult to ensure the right dead time (even with an inner control loop) over all conditions (including transients), probably we should stick to the wise principle

"better safe than sorry"

by assigning the dead time long enough for the worst case condition.

Joseph's concern of the loss of duty cycle due to long dead times is unfounded, because dead times might account for only a few percents of a switching period at most.
11-21-2013 02:17 PM
Top #6
Darrell Hambley P.E.
11-21-2013 02:17 PM
The effect of too long of a dead time can result in a large voltage across the capacitance, not just a few volts. There are 2 worse-case-conditions: 1. Too little dead time and 2. Too much dead time. Over the range of min to max load this expands to 4 corners of worse-case-conditions.
11-21-2013 04:32 PM
Top #7
Ray Ridley
11-21-2013 04:32 PM
it keeps coming back to this in power electronics: turn on your switches, and turn them off at just the right time. What is a shock to newcomers to the field is how precise the timing has to be to get it just right, keep things cool, and stop the circuit blowing up.

The ns timing has kept this field challenging and will continue to do so.

Add to that the punishment factor of the degree of destruction when things are not done right, it tends to keep power electronics engineers pretty down to earth and honest.

Trying to identify every scenario of how things can go wrong, especially in more complex topologies and technologies, is going to keep engineers busy for quite some time.
11-21-2013 06:33 PM
Top #8
Mau Pham
11-21-2013 06:33 PM
Darrell, Ray:

If the worst case corners are such that setting the dead time to the long worst case will cause the Drain-Source Voltage of a Power Transistor to sometimes reach several tens of percents of the supply voltage (in the short worst case), then you probably need to design some kind of tracking (with temperature, with supply voltage etc.) to narrow the range of worst cases.

But inserting an inner control loop just to optimize the dead time might interfere with the outer (i.e. regular) control loop, and still might not be able to follow transients.

Anyway, setting a fix dead time which errs in the too long direction will definitely

"stop the circuit blowing up" as Ray worries.

and makes "Power Electronics" easy enough for cavemen as well as for experienced power electronic engineers (who are increasingly hard to find)
Reply to Thread